X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=src%2FProtocol%2FFeed.php;h=32ff61756df3eb36de1a6955b066edd2d17864bb;hb=14fde5dc9b1915392601fb94efc6224c01f2b216;hp=203a2e83565d27396dcb4dc7486e17e8a5bab5a8;hpb=6d19c780d9b26a02f883563c6552f1a4f999d262;p=friendica.git diff --git a/src/Protocol/Feed.php b/src/Protocol/Feed.php index 203a2e8356..32ff61756d 100644 --- a/src/Protocol/Feed.php +++ b/src/Protocol/Feed.php @@ -6,16 +6,18 @@ */ namespace Friendica\Protocol; -use Friendica\Database\DBM; +use DOMDocument; +use DOMXPath; +use Friendica\Content\Text\HTML; +use Friendica\Core\Logger; +use Friendica\Core\Protocol; use Friendica\Core\System; +use Friendica\Database\DBA; use Friendica\Model\Item; use Friendica\Util\Network; -use dba; -use DOMDocument; -use DOMXPath; +use Friendica\Util\XML; require_once 'include/dba.php'; -require_once 'include/html2bbcode.php'; require_once 'include/items.php'; /** @@ -39,12 +41,12 @@ class Feed { $a = get_app(); if (!$simulate) { - logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG); + Logger::log("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG); } else { - logger("Test Atom/RSS feed", LOGGER_DEBUG); + Logger::log("Test Atom/RSS feed", LOGGER_DEBUG); } - if ($xml == "") { - logger('XML is empty.', LOGGER_DEBUG); + if (empty($xml)) { + Logger::log('XML is empty.', LOGGER_DEBUG); return; } @@ -68,108 +70,106 @@ class Feed { $xpath->registerNamespace('poco', NAMESPACE_POCO); $author = []; + $entries = null; // Is it RDF? if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) { - $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue; - $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue; + $author["author-link"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:link/text()'); + $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:title/text()'); - if ($author["author-name"] == "") { - $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue; + if (empty($author["author-name"])) { + $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:description/text()'); } $entries = $xpath->query('/rdf:RDF/rss:item'); } // Is it Atom? if ($xpath->query('/atom:feed')->length > 0) { - $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes; + $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']"); if (is_object($alternate)) { - foreach ($alternate AS $attributes) { - if ($attributes->name == "href") { - $author["author-link"] = $attributes->textContent; + foreach ($alternate AS $attribute) { + if ($attribute->name == "href") { + $author["author-link"] = $attribute->textContent; } } } - if ($author["author-link"] == "") { - $author["author-link"] = $author["author-id"]; - } - if ($author["author-link"] == "") { - $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes; + if (empty($author["author-link"])) { + $self = XML::getFirstAttributes($xpath, "atom:link[@rel='self']"); if (is_object($self)) { - foreach ($self AS $attributes) { - if ($attributes->name == "href") { - $author["author-link"] = $attributes->textContent; + foreach ($self AS $attribute) { + if ($attribute->name == "href") { + $author["author-link"] = $attribute->textContent; } } } } - if ($author["author-link"] == "") { - $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue; + if (empty($author["author-link"])) { + $author["author-link"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:id/text()'); } - $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue; + $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:logo/text()'); - $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue; + $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:title/text()'); - if ($author["author-name"] == "") { - $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue; + if (empty($author["author-name"])) { + $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:subtitle/text()'); } - if ($author["author-name"] == "") { - $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue; + if (empty($author["author-name"])) { + $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:name/text()'); } - $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue; + $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:displayName/text()'); if ($value != "") { $author["author-name"] = $value; } if ($simulate) { - $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue; + $author["author-id"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:uri/text()'); - $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue; + $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:preferredUsername/text()'); if ($value != "") { $author["author-nick"] = $value; } - $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()')->item(0)->nodeValue; + $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:address/poco:formatted/text()'); if ($value != "") { $author["author-location"] = $value; } - $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue; + $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:note/text()'); if ($value != "") { $author["author-about"] = $value; } - $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes; + $avatar = XML::getFirstAttributes($xpath, "atom:author/atom:link[@rel='avatar']"); if (is_object($avatar)) { - foreach ($avatar AS $attributes) { - if ($attributes->name == "href") { - $author["author-avatar"] = $attributes->textContent; + foreach ($avatar AS $attribute) { + if ($attribute->name == "href") { + $author["author-avatar"] = $attribute->textContent; } } } } - $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue; + $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:updated/text()'); - $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue; + $author["app"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:generator/text()'); $entries = $xpath->query('/atom:feed/atom:entry'); } // Is it RSS? if ($xpath->query('/rss/channel')->length > 0) { - $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue; + $author["author-link"] = XML::getFirstNodeValue($xpath, '/rss/channel/link/text()'); - $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue; - $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue; + $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/title/text()'); + $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/rss/channel/image/url/text()'); - if ($author["author-name"] == "") { - $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue; + if (empty($author["author-name"])) { + $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/copyright/text()'); } - if ($author["author-name"] == "") { - $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue; + if (empty($author["author-name"])) { + $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/description/text()'); } - $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue; + $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/rss/channel/pubDate/text()'); - $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue; + $author["app"] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()'); $entries = $xpath->query('/rss/channel/item'); } @@ -177,7 +177,7 @@ class Feed { if (!$simulate) { $author["author-link"] = $contact["url"]; - if ($author["author-name"] == "") { + if (empty($author["author-name"])) { $author["author-name"] = $contact["name"]; } $author["author-avatar"] = $contact["thumb"]; @@ -189,8 +189,7 @@ class Feed { $header = []; $header["uid"] = $importer["uid"]; - $header["network"] = NETWORK_FEED; - $header["type"] = "remote"; + $header["network"] = Protocol::FEED; $header["wall"] = 0; $header["origin"] = 0; $header["gravity"] = GRAVITY_PARENT; @@ -201,44 +200,41 @@ class Feed { $header["contact-id"] = $contact["id"]; if (!is_object($entries)) { - logger("There are no entries in this feed.", LOGGER_DEBUG); + Logger::log("There are no entries in this feed.", LOGGER_DEBUG); return; } $items = []; + // Importing older entries first + for($i = $entries->length - 1; $i >= 0;--$i) { + $entry = $entries->item($i); - $entrylist = []; - - foreach ($entries AS $entry) { - $entrylist[] = $entry; - } - foreach (array_reverse($entrylist) AS $entry) { $item = array_merge($header, $author); - $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes; + $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']", $entry); if (!is_object($alternate)) { - $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes; + $alternate = XML::getFirstAttributes($xpath, "atom:link", $entry); } if (is_object($alternate)) { - foreach ($alternate AS $attributes) { - if ($attributes->name == "href") { - $item["plink"] = $attributes->textContent; + foreach ($alternate AS $attribute) { + if ($attribute->name == "href") { + $item["plink"] = $attribute->textContent; } } } - if ($item["plink"] == "") { - $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue; + if (empty($item["plink"])) { + $item["plink"] = XML::getFirstNodeValue($xpath, 'link/text()', $entry); } - if ($item["plink"] == "") { - $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue; + if (empty($item["plink"])) { + $item["plink"] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry); } - $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue; + $item["uri"] = XML::getFirstNodeValue($xpath, 'atom:id/text()', $entry); - if ($item["uri"] == "") { - $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue; + if (empty($item["uri"])) { + $item["uri"] = XML::getFirstNodeValue($xpath, 'guid/text()', $entry); } - if ($item["uri"] == "") { + if (empty($item["uri"])) { $item["uri"] = $item["plink"]; } @@ -250,53 +246,58 @@ class Feed { if (!$simulate) { $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)", - $importer["uid"], $item["uri"], NETWORK_FEED, NETWORK_DFRN]; - $previous = dba::selectFirst('item', ['id'], $condition); - if (DBM::is_result($previous)) { - logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], LOGGER_DEBUG); + $importer["uid"], $item["uri"], Protocol::FEED, Protocol::DFRN]; + $previous = Item::selectFirst(['id'], $condition); + if (DBA::isResult($previous)) { + Logger::log("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], LOGGER_DEBUG); continue; } } - $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue; + $item["title"] = XML::getFirstNodeValue($xpath, 'atom:title/text()', $entry); - if ($item["title"] == "") { - $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue; + if (empty($item["title"])) { + $item["title"] = XML::getFirstNodeValue($xpath, 'title/text()', $entry); } - if ($item["title"] == "") { - $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue; + if (empty($item["title"])) { + $item["title"] = XML::getFirstNodeValue($xpath, 'rss:title/text()', $entry); } - $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue; + $published = XML::getFirstNodeValue($xpath, 'atom:published/text()', $entry); - if ($published == "") { - $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue; + if (empty($published)) { + $published = XML::getFirstNodeValue($xpath, 'pubDate/text()', $entry); } - if ($published == "") { - $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue; + if (empty($published)) { + $published = XML::getFirstNodeValue($xpath, 'dc:date/text()', $entry); } - $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue; + $updated = XML::getFirstNodeValue($xpath, 'atom:updated/text()', $entry); - if ($updated == "") { + if (empty($updated) && !empty($published)) { $updated = $published; } + + if (empty($published) && !empty($updated)) { + $published = $updated; + } + if ($published != "") { $item["created"] = $published; } if ($updated != "") { $item["edited"] = $updated; } - $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue; + $creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry); - if ($creator == "") { - $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue; + if (empty($creator)) { + $creator = XML::getFirstNodeValue($xpath, 'atom:author/atom:name/text()', $entry); } - if ($creator == "") { - $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue; + if (empty($creator)) { + $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry); } if ($creator != "") { $item["author-name"] = $creator; } - $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue; + $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry); if ($creator != "") { $item["author-name"] = $creator; @@ -308,24 +309,26 @@ class Feed { $attachments = []; - $enclosures = $xpath->query("enclosure", $entry); + $enclosures = $xpath->query("enclosure|atom:link[@rel='enclosure']", $entry); foreach ($enclosures AS $enclosure) { $href = ""; $length = ""; $type = ""; $title = ""; - foreach ($enclosure->attributes AS $attributes) { - if ($attributes->name == "url") { - $href = $attributes->textContent; - } elseif ($attributes->name == "length") { - $length = $attributes->textContent; - } elseif ($attributes->name == "type") { - $type = $attributes->textContent; + foreach ($enclosure->attributes AS $attribute) { + if (in_array($attribute->name, ["url", "href"])) { + $href = $attribute->textContent; + } elseif ($attribute->name == "length") { + $length = $attribute->textContent; + } elseif ($attribute->name == "type") { + $type = $attribute->textContent; } } - if (strlen($item["attach"])) { + if (!empty($item["attach"])) { $item["attach"] .= ','; + } else { + $item["attach"] = ''; } $attachments[] = ["link" => $href, "type" => $type, "length" => $length]; @@ -345,16 +348,16 @@ class Feed { $tags .= $taglink; } - $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue); + $body = trim(XML::getFirstNodeValue($xpath, 'atom:content/text()', $entry)); - if ($body == "") { - $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue); + if (empty($body)) { + $body = trim(XML::getFirstNodeValue($xpath, 'content:encoded/text()', $entry)); } - if ($body == "") { - $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue); + if (empty($body)) { + $body = trim(XML::getFirstNodeValue($xpath, 'description/text()', $entry)); } - if ($body == "") { - $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue); + if (empty($body)) { + $body = trim(XML::getFirstNodeValue($xpath, 'atom:summary/text()', $entry)); } // remove the content of the title if it is identically to the body @@ -362,16 +365,15 @@ class Feed { if (self::titleIsBody($item["title"], $body)) { $item["title"] = ""; } - $item["body"] = html2bbcode($body, $basepath); + $item["body"] = HTML::toBBCode($body, $basepath); if (($item["body"] == '') && ($item["title"] != '')) { $item["body"] = $item["title"]; $item["title"] = ''; } + $preview = ''; if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) { - $preview = ""; - // Handle enclosures and treat them as preview picture foreach ($attachments AS $attachment) { if ($attachment["type"] == "image/jpeg") { @@ -402,7 +404,7 @@ class Feed { // We always strip the title since it will be added in the page information $item["title"] = ""; $item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]); - $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]); + $item["tag"] = add_page_keywords($item["plink"], $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]); $item["object-type"] = ACTIVITY_OBJ_BOOKMARK; unset($item["attach"]); } else { @@ -410,7 +412,8 @@ class Feed { if (!empty($tags)) { $item["tag"] = $tags; } else { - $item["tag"] = add_page_keywords($item["plink"], false, $preview, true, $contact["ffi_keyword_blacklist"]); + // @todo $preview is never set in this case, is it intended? - @MrPetovan 2018-02-13 + $item["tag"] = add_page_keywords($item["plink"], $preview, true, $contact["ffi_keyword_blacklist"]); } $item["body"] .= "\n".$item['tag']; } @@ -421,21 +424,24 @@ class Feed { } if (!$simulate) { - logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG); + Logger::log("Stored feed: ".print_r($item, true), LOGGER_DEBUG); $notify = Item::isRemoteSelf($contact, $item); // Distributed items should have a well formatted URI. // Additionally we have to avoid conflicts with identical URI between imported feeds and these items. if ($notify) { - $item['guid'] = Item::guidFromUri($orig_plink, $a->get_hostname()); + $item['guid'] = Item::guidFromUri($orig_plink, $a->getHostName()); unset($item['uri']); unset($item['parent-uri']); + + // Set the delivery priority for "remote self" to "medium" + $notify = PRIORITY_MEDIUM; } $id = Item::insert($item, false, $notify); - logger("Feed for contact ".$contact["url"]." stored under id ".$id); + Logger::log("Feed for contact ".$contact["url"]." stored under id ".$id); } else { $items[] = $item; }